---
title: "Maximizing Your Website: Client Question Analysis"
output: 
  flexdashboard::flex_dashboard:
    theme:
      version: 4
      bootswatch: minty
      navbar-bg: "#005E86"
    orientation: columns
    source_code: embed
---

<style>
.chart-title {  /* chart_title  */
   font-size: 20px;
   font-family: Helvetica;
  }
body{ /* Normal  */
      font-size: 18px;
  }
</style>

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE)
```

```{r packages, echo = FALSE}
pacman::p_load(shiny, readxl, tm, ggplot2, plotly, readr, dplyr, stringr,
               SnowballC, wordcloud, RColorBrewer, wordcloud2, syuzhet,
               tidyverse, tidytext, igraph, ggraph, widyr)
```

Cover Page
===

Column {data-width=500}
---

### Introduction

An online platform is offered by the <b> American Bar Association </b> to provide *pro bono* legal assistance (free of charge) across the United States. The platform enables individuals who meet the state-specific criteria of low-income status to submit legal queries and receive guidance from volunteer attorneys.

The goal for ABA is:

➼ Anticipate the sort of legal questions that arise so they can better prepare volunteers.

➼ Better understand how and when to recruit lawyers with specific expertise.

➼ Develop a stronger dialogue with state partners on general trends they're seeing.

### 

```{r pic1, out.width="600px"}
knitr::include_graphics("G:/My Drive/Datafest/pic1.png")
```

Column {data-length=500}
---

### Our Approach

We focused on the clientele with the smallest amount of interaction; no responses to their questions. It is from this subset we extended our analysis. Storylines we explored are:

➼ What is the most prominent category of these unanswered questions?

➼ What are the most commonly used *buzz* words/phrases in this category?

➼ Clients that use these *buzz* words, is there any underlying connection related to demographics?

➼ What separates/differs from questions that get answered compared to those that go unanswered?

Using <b> Text Mining </b> and standard Exploratory Data Analysis techniques, this study will uncover deeper interactions between clients and the questions they ask. 

### Motivation

How would an analysis of this nature benefit ABA? Well, the overarching goal is to provide information to ABA on the...

➼ *types* of questions that get asked.

➼ *who* exactly is asking these questions.

➼ *what* kinds of lawyers are needed to answer said questions.

In tandem with this study, we can provide exactly that.

Category Analysis
===

Column {data-width=300}
-----------------------------------------------------------------------

### Analysis

➼ After analyzing the "Category" variable to determine the types of questions asked, we discovered that the Family and Children category was the most frequently chosen by clients.

➼ Related subcategories include: Family/Divorce/Custody, Wills/Trusts/Estates, Paternity, Adoption, and Child Support. 

➼ Presently, Domestic Relation Lawyers are recognized as an essential legal service that individuals may require at some point in their lives.([The National Trial Lawyers](https://thenationaltriallawyers.org/article/the-top-10-different-types-of-lawyers-you-might-need/))

Column {data-width=650}
-----------------------------------------------------------------------

### Category Bar Chart

```{r pic2, out.width="600px"}
knitr::include_graphics("G:/My Drive/Datafest/pic2.png")
```

Text Mining
===

Column {data-width=400}
-----------------------------------------------------------------------

### Word Cloud

We produced two word clouds to compare the buzz words involved in the family cases for both the answered and unanswered questions. The results found are:

➼ The key words we found in both word clouds include "child", "custody", and "support", which suggests that more attorneys who specialize in child support or custody battles are needed in the program.

➼ We also noticed that there are very similar keywords between the two word clouds, indicating that the wording of the questions does not seem to be the reason for unanswered questions.

➼ However, further analysis will be required to determine if there are any significant differences and whether more attorneys should be recruited for the program.

➼ The most common issues handled at family court include: Marriage Dissolution (<b><font color="#0089B5">divorce</font></b>), Paternity/Child Custody (<b><font color="#05B3E4">custody</font></b>), Domestic Violence, Name Changes(<b><font color="#05B3E4">name</font></b>),Guardianship, Adoptions ([Family Law Self Help Center](https://www.familylawselfhelpcenter.org/self-help/getting-started/court-basics/types-of-cases))

```{css color tabs, echo = FALSE}
/* Set font color of inactive tab to green */
.nav-tabs-custom .nav-tabs > li > a 
  {
    color: #0089B5;
  } 

/* Set font color of active tab to red */
.nav-tabs-custom .nav-tabs > li.active > a 
  {
    color: #F88E28;
  } 

/* To set color on hover */
.nav-tabs-custom .nav-tabs > li.active > a:hover 
  {
    color: grey;
  }

<style type="text/css"> .sidebar 
  { 
    overflow: auto; 
  } 
</style>
```

Column {data-width=550, .tabset}
-----------------------------------------------------------------------

### Answered

```{r pic3, out.width="600px"}
knitr::include_graphics("G:/My Drive/Datafest/pic3.png")
```

### Unanswered

```{r pic4, out.width="600px"}
knitr::include_graphics("G:/My Drive/Datafest/pic4.png")
```

Map
===

Column {data-width=300}
-----------------------------------------------------------------------

### Analysis

➼ States colored in grey have no observations within the "Family and Children" category, meaning no questions originated from them. It is unlikely that these states have zero representation from the Family and Children category so there must be some other factor that causes this. 

➼ The states with the darkest colors have the most unanswered questions: Pennsylvania, Georgia, and Kansas. 

➼ A logical extension from the analysis of darker states is the lack of licensed lawyers within those states. Attorneys are restricted to serving clients who have posted from the same state in which they hold a valid license. However, to delve deeper into this issue, additional data is necessary.


Column {data-width=650}
-----------------------------------------------------------------------

### Map

```{r pic5, out.width="600px"}
knitr::include_graphics("G:/My Drive/Datafest/pic5.png")
```

Conclusion
===

Column {data-width=700}
-----------------------------------------------------------------------

### Conclusion

<b>Implications</b>: 

➼ From our analysis on text and category, we found no discrepancy between the way the questions that went unanswered versus the questions that were answered were phrased and asked. Due to the grammatical similarities, we believe that the burden of a question not being answered does not fall on the client, but rather some external factor. (pro bono is free work that is not required for a lawyer and is also on their on time)

➼ To increase outreach, our team believes dark colored states are the best places to start. Regarding lawyers (specifically, family relations) licensed in these states, a possible solution could include to push them to pursue more pro bono opportunities. 

➼ ABA should reach out to state representatives regarding internet access for any of the grey states (the states in which there was no observations for family-related questions in general). Those who cannot afford internet are under the same demographic of those who probably cannot afford a lawyer, but are unable to use the pro bono services. 


<b>Limitations</b>:

➼ Regarding grey states, we would have enjoyed to have data involving internet access. With this, we could have analyzed for ourselves if internet access was the driving factor for grey states. But, as is, this stands as an implication for ABA.

➼ We would have liked to have more data regarding the education levels of the clients. This would have allowed us to perform further analysis to see if education had a relationship with questions going answered or unanswered.

➼ What kind of marketing strategies are in place for the website? How much are they spending on getting this resource into the hands of the people?


Column {data-width=250}
-----------------------------------------------------------------------

### Acknowledgments

A huge thank you to ASA for putting on and to Miami University for hosting DataFest 2023! 

We would also like to thank Dr. Tessa Chen and Dr. Matthew Wascher for advising and caring for our team. We would not have been able to be here without them!


Appendix
===

Row 
-----------------------------------------------------------------------
  
```{r app1, out.width="600px"}
knitr::include_graphics("G:/My Drive/Datafest/pic6.png")

knitr::include_graphics("G:/My Drive/Datafest/pic7.png")
```

Row
-----------------------------------------------------------------------

```{r app2, out.width="600px"}
knitr::include_graphics("G:/My Drive/Datafest/pic8.png")

knitr::include_graphics("G:/My Drive/Datafest/pic9.png")
```